Remove common.PORT usage from `parallel` tests #12473

thelostone-mc · 2017-04-17T18:28:17Z

Refs: #12376

Tests Updated:

test/parallel/test-net-better-error-messages-port-hostname.js
test/parallel/test-net-socket-destroy-twice.js
test/parallel/test-net-options-lookup.js
test/parallel/test-net-connect-immediate-destroy.js
test/parallel/test-net-localerror.js
test/parallel/test-net-connect-handle-econnrefused.js

Tests ported from test/parallel to test/sequential:

test/sequential/test-net-listen-shared-ports.js
test/sequential/test-net-better-error-messages-port.js

Analyzed and continuing to use common.PORT

test/parallel/test-net-connect-immediate-finish.js

Checklist

make -j4 test (UNIX), or vcbuild test (Windows) passes
tests and/or benchmarks are included
commit message follows commit guidelines

Affected core subsystem(s)

test

thelostone-mc · 2017-04-17T18:29:42Z

First-time contributor. (Noob level)
Anything I'm missing out on?

Thanks for pushing me to get into this!
@thefourtheye

vsemozhetbyt · 2017-04-17T18:54:44Z

CI: https://ci.nodejs.org/job/node-test-pull-request/7455/

thefourtheye · 2017-04-18T01:47:24Z

Good job @adityaanandmc 👏

thefourtheye · 2017-04-18T05:59:55Z

@adityaanandmc Changes are all good. If you could make sure if the commits follow these guidelines it would be great?

thelostone-mc · 2017-04-18T09:21:06Z

@thefourtheye Do the changes look alright ?

thelostone-mc · 2017-04-18T17:34:57Z

@thefourtheye @cjihrig
Are we allowed to do this? Create a server and kill it simply to get a free port.
Is there a way to just get an unused port ?

cjihrig · 2017-04-18T17:38:04Z

Are we allowed to do this? Create a server and kill it simply to get a free port.

That would probably introduce a race condition. Once you close the server to release the port, something else can claim that port.

thelostone-mc · 2017-04-19T07:04:18Z

@cjihrig What would be the ideal way to get this done? I was aware of the race condition and hence a little stuck. My commit ( code below ) will fail if the race condition is met.
How do I secure a free port and listen to it ensuring that a server hasn't been created in the gap?

const net = require('net');
const assert = require('assert');

//const c = net.createConnection(common.PORT);
const server = net.createServer();
server.listen(0);
const port = server.address().port;
server.close();

const c = net.createConnection(port);

c.on('connect', common.mustNotCall());

c.on('error', common.mustCall(function(e) {
    assert.strictEqual(e.code, 'ECONNREFUSED');
    //assert.strictEqual(e.port, common.PORT);
    assert.strictEqual(e.port, port);
    assert.strictEqual(e.address, '127.0.0.1');
}));

thefourtheye · 2017-04-19T07:34:54Z

@adityaanandmc wrote

How do I secure a free port and listen to it ensuring that a server hasn't been created in the gap?

Although the possibility of this happening is less, it is still possible and we cannot avoid it.

santigimeno · 2017-04-19T07:42:58Z

@adityaanandmc what about doing something like this:

...
const server = net.createServer();
server.listen(0);
const port = server.address().port;
const c = net.createConnection(port);
server.close();
...

thelostone-mc · 2017-04-19T08:39:06Z

@santigimeno Now i feel like a idiot :P Thanks for that!! :D

santigimeno · 2017-04-19T16:38:06Z

@adityaanandmc could you squash the commits into one? Thanks!

jasnell · 2017-04-19T17:03:19Z

CI: https://ci.nodejs.org/job/node-test-pull-request/7523/

thelostone-mc · 2017-04-19T18:18:29Z

@santigimeno will squash them!!
Looks like there are a few failures test-net-connect-immediate-finish

I can't seem to understand the failures. I thought about doing this, but it didn't seem like the right thing to do.

const server = net.createServer();
server.listen(0);
const port = server.address().port;
server.close();

const client = net.connect({host: '***', port: port});

santigimeno · 2017-04-19T19:36:12Z

@adityaanandmc looking at the changes again it seems that we're trying to connect() to the 0 port. I don't think that makes sense: the port 0 makes sense when calling bind() or listen() not connect(). What I think it could be done is create a helper function in common.js (connectToRandomPort for example) to reserve a port and connect, similar to the code you were proposing.

thelostone-mc · 2017-04-20T16:00:37Z

@santigimeno @thefourtheye I did that and it passed locally but at times it fails with the following error!
It started only after i had touched this file test-net-connect-immediate-finish.
Not able to figure out why it's happening.

Path: parallel/test-domain-no-error-handler-abort-on-uncaught-0
Command: out/Release/node /Users/adianand/Labs/node/test/parallel/test-domain-no-error-handler-abort-on-uncaught-0.js
--- TIMEOUT ---
=== release test-domain-no-error-handler-abort-on-uncaught-1 ===
Path: parallel/test-domain-no-error-handler-abort-on-uncaught-1
Command: out/Release/node /Users/adianand/Labs/node/test/parallel/test-domain-no-error-handler-abort-on-uncaught-1.js
--- TIMEOUT ---
=== release test-domain-no-error-handler-abort-on-uncaught-2 ===
Path: parallel/test-domain-no-error-handler-abort-on-uncaught-2
Command: out/Release/node /Users/adianand/Labs/node/test/parallel/test-domain-no-error-handler-abort-on-uncaught-2.js
--- TIMEOUT ---
=== release test-domain-no-error-handler-abort-on-uncaught-3 ===
Path: parallel/test-domain-no-error-handler-abort-on-uncaught-3
Command: out/Release/node /Users/adianand/Labs/node/test/parallel/test-domain-no-error-handler-abort-on-uncaught-3.js
--- TIMEOUT ---
=== release test-domain-no-error-handler-abort-on-uncaught-4 ===
Path: parallel/test-domain-no-error-handler-abort-on-uncaught-4
Command: out/Release/node /Users/adianand/Labs/node/test/parallel/test-domain-no-error-handler-abort-on-uncaught-4.js
--- TIMEOUT ---

Sorry for flooding you with doubts. Just a lil confused for a first-time contributor on a project of this scale.

addaleax · 2017-05-06T16:00:39Z

I think the actual error in the output you linked to is this:

not ok 357 sequential/test-benchmark-child-process
  ---
  duration_ms: 60.190
  severity: fail
  stack: |-
    timeout

That is unrelated, and was tracked at #12817 (but should be fixed now).

Either way, it looks like this PR has a merge conflict now so it might need to rebase – is that something you can do?

santigimeno · 2017-05-06T16:04:10Z

@adityaanandmc, what @addaleax said.

BTW, could you address this nit? Thanks!

thelostone-mc · 2017-05-06T16:23:50Z

Sure thing !! Will do that in a bit though. ( heading out to watch Guardians of the Galaxy ) :D

thelostone-mc · 2017-05-07T08:00:38Z

@addaleax @santigimeno Rebased!! Thanks for that. Can a CI job be triggered?

gibfahn · 2017-05-07T14:47:42Z

CI: https://ci.nodejs.org/job/node-test-commit/9717/

santigimeno · 2017-05-07T18:21:55Z

@adityaanandmc sorry to bring this up again, but it would be great if you could address this. Anyway, the PR as it is LGTM. Thanks!

thelostone-mc · 2017-05-07T18:52:10Z

@santigimeno That's alright. I did not see that comment until now. Sorry about that!
Added comments ! Do they look alright ?

Updated:

Thanks @santigimeno @thefourtheye @Trott @cjihrig !!

santigimeno · 2017-05-07T20:53:57Z

Added comments ! Do they look alright ?

Yes, LGTM. I think another in test-net-localerror.js could be useful. Thanks!

thelostone-mc · 2017-05-08T04:17:39Z

@santigimeno Done!

thefourtheye · 2017-05-08T05:29:22Z

test/parallel/test-net-better-error-messages-port.js

  assert.strictEqual(e.address, '127.0.0.1');
 }));
+server.close();


Okay, server.close is actually asynchronous. Perhaps we should run this test in the close event of server.

Okay, server.close is actually asynchronous

You're right, though the close() is executed syncronously at least on Unixes, but that's an implementation detail.

Perhaps we should run this test in the close event of server.

I think that's not going to work either, because there's going to be a window when the port is not assigned and could be grabbed by some other process/test.

So probably the only way to make this work without race conditions is moving it to sequential. What do you think @thefourtheye ?

@santigimeno close() is actually executed synchronously, but by the time the connection attempt is made (which happens asynchronously), the port is still up for grab, right? Perhaps I am overthinking 🤔

I am afraid sequential also will not solve this. common.PORT could still have been assigned to other processes.

Perhaps we can live with the current change...

@santigimeno close() is actually executed synchronously, but by the time the connection attempt is made (which happens asynchronously), the port is still up for grab, right? Perhaps I am overthinking 🤔

Yes, I thought of that too. And I think I'm also overthinking 😆 .

I am afraid sequential also will not solve this. common.PORT could still have been assigned to other processes.

Right, but not to other tests in the same run though.

Perhaps we can live with the current change...

Yeah, probably they'll never fail.

Since we've reached a moo point. Would it be alright if I followed @thefourtheye approach and change the code accordingly ?

@adityaanandmc No, move this to sequential, without the changes.

Remove common.PORT from, test-net-connect-immediate-destroy, test-net-options-lookup, test-net-connect-local-error, test-net-connect-handle-econnrefused, test-net-socket-destroy-twice, test-net-better-error-messages-port-hostname, test-net-localerror, to reduce possibility that a dynamic port used in another test will collide with common.PORT. Moved test-net-listen-shared-ports, test-net-better-error-messages-port from tests/parallel to test/sequential Refs: nodejs#12376

thelostone-mc · 2017-05-08T16:58:21Z

@thefourtheye Done. CI Build trigger necessary ?

thefourtheye · 2017-05-09T17:20:45Z

CI Run: https://ci.nodejs.org/job/node-test-pull-request/7965/

Remove common.PORT from, test-net-connect-immediate-destroy, test-net-options-lookup, test-net-connect-local-error, test-net-connect-handle-econnrefused, test-net-socket-destroy-twice, test-net-better-error-messages-port-hostname, test-net-localerror, to reduce possibility that a dynamic port used in another test will collide with common.PORT. Moved test-net-listen-shared-ports, test-net-better-error-messages-port from tests/parallel to test/sequential Refs: #12376 PR-URL: #12473 Reviewed-By: James M Snell <[email protected]> Reviewed-By: Colin Ihrig <[email protected]> Reviewed-By: Yuta Hiroto <[email protected]> Reviewed-By: Santiago Gimeno <[email protected]> Reviewed-By: Sakthipriyan Vairamani <[email protected]>

jasnell · 2017-05-09T20:28:00Z

Landed in 94eed0f

Remove common.PORT from, test-net-connect-immediate-destroy, test-net-options-lookup, test-net-connect-local-error, test-net-connect-handle-econnrefused, test-net-socket-destroy-twice, test-net-better-error-messages-port-hostname, test-net-localerror, to reduce possibility that a dynamic port used in another test will collide with common.PORT. Moved test-net-listen-shared-ports, test-net-better-error-messages-port from tests/parallel to test/sequential Refs: nodejs#12376 PR-URL: nodejs#12473 Reviewed-By: James M Snell <[email protected]> Reviewed-By: Colin Ihrig <[email protected]> Reviewed-By: Yuta Hiroto <[email protected]> Reviewed-By: Santiago Gimeno <[email protected]> Reviewed-By: Sakthipriyan Vairamani <[email protected]>

Remove common.PORT from, test-net-connect-immediate-destroy, test-net-options-lookup, test-net-connect-local-error, test-net-connect-handle-econnrefused, test-net-socket-destroy-twice, test-net-better-error-messages-port-hostname, test-net-localerror, to reduce possibility that a dynamic port used in another test will collide with common.PORT. Moved test-net-listen-shared-ports, test-net-better-error-messages-port from tests/parallel to test/sequential Refs: #12376 PR-URL: #12473 Reviewed-By: James M Snell <[email protected]> Reviewed-By: Colin Ihrig <[email protected]> Reviewed-By: Yuta Hiroto <[email protected]> Reviewed-By: Santiago Gimeno <[email protected]> Reviewed-By: Sakthipriyan Vairamani <[email protected]>

nodejs-github-bot added the test Issues and PRs related to the tests. label Apr 17, 2017

cjihrig approved these changes Apr 17, 2017

View reviewed changes

jasnell approved these changes Apr 17, 2017

View reviewed changes

santigimeno approved these changes Apr 17, 2017

View reviewed changes

mscdex added the net Issues and PRs related to the net subsystem. label Apr 17, 2017

hiroppy approved these changes Apr 18, 2017

View reviewed changes

thefourtheye approved these changes Apr 18, 2017

View reviewed changes

Trott mentioned this pull request Apr 18, 2017

Remove common.PORT usage from parallel tests #12376

Closed

64 tasks

thelostone-mc force-pushed the master branch from 5a39c4b to 5b312a4 Compare April 18, 2017 06:50

thelostone-mc changed the title ~~Removing common.PORT usage from parallel tests~~ Remove common.PORT usage from parallel tests Apr 18, 2017

thelostone-mc force-pushed the master branch from 5b312a4 to 28fe043 Compare April 18, 2017 07:38

thelostone-mc force-pushed the master branch from a6e8e5e to e8b3179 Compare April 19, 2017 08:34

thelostone-mc force-pushed the master branch 2 times, most recently from 21c212d to 582ef62 Compare April 20, 2017 15:05

thelostone-mc force-pushed the master branch from 135cee1 to 3e82e88 Compare May 7, 2017 05:58

thelostone-mc force-pushed the master branch from 3e82e88 to b2bb260 Compare May 7, 2017 18:47

thelostone-mc force-pushed the master branch from b2bb260 to 9e3b159 Compare May 8, 2017 04:17

thefourtheye reviewed May 8, 2017

View reviewed changes

thelostone-mc force-pushed the master branch from 9e3b159 to ee84852 Compare May 8, 2017 16:54

thelostone-mc force-pushed the master branch from ee84852 to f5aead0 Compare May 8, 2017 16:57

thefourtheye approved these changes May 9, 2017

View reviewed changes

jasnell closed this May 9, 2017

jasnell mentioned this pull request May 11, 2017

8.0.0 Release Proposal #12220

Closed

gibfahn mentioned this pull request Jun 15, 2017

Auditing for 6.11.1 nodejs/Release#230

Closed

3 tasks

gibfahn added the land-on-v6.x label Jun 20, 2017

MylesBorins mentioned this pull request Jul 18, 2017

v6.11.2 proposal #14356

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Remove common.PORT usage from `parallel` tests #12473

Remove common.PORT usage from `parallel` tests #12473

thelostone-mc commented Apr 17, 2017 •

edited

Loading

thelostone-mc commented Apr 17, 2017 •

edited

Loading

vsemozhetbyt commented Apr 17, 2017

thefourtheye commented Apr 18, 2017 •

edited

Loading

thefourtheye commented Apr 18, 2017

thelostone-mc commented Apr 18, 2017

thelostone-mc commented Apr 18, 2017

cjihrig commented Apr 18, 2017

thelostone-mc commented Apr 19, 2017

thefourtheye commented Apr 19, 2017

santigimeno commented Apr 19, 2017

thelostone-mc commented Apr 19, 2017 •

edited

Loading

santigimeno commented Apr 19, 2017

jasnell commented Apr 19, 2017

thelostone-mc commented Apr 19, 2017

santigimeno commented Apr 19, 2017

thelostone-mc commented Apr 20, 2017 •

edited

Loading

addaleax commented May 6, 2017

santigimeno commented May 6, 2017 •

edited

Loading

thelostone-mc commented May 6, 2017

thelostone-mc commented May 7, 2017 •

edited

Loading

gibfahn commented May 7, 2017

santigimeno commented May 7, 2017

thelostone-mc commented May 7, 2017

santigimeno commented May 7, 2017

thelostone-mc commented May 8, 2017

thefourtheye May 8, 2017

santigimeno May 8, 2017 •

edited

Loading

thefourtheye May 8, 2017

santigimeno May 8, 2017

thelostone-mc May 8, 2017

thefourtheye May 8, 2017

thelostone-mc commented May 8, 2017

thefourtheye commented May 9, 2017

jasnell commented May 9, 2017

Remove common.PORT usage from parallel tests #12473

Remove common.PORT usage from parallel tests #12473

Conversation

thelostone-mc commented Apr 17, 2017 • edited Loading

Checklist

Affected core subsystem(s)

thelostone-mc commented Apr 17, 2017 • edited Loading

vsemozhetbyt commented Apr 17, 2017

thefourtheye commented Apr 18, 2017 • edited Loading

thefourtheye commented Apr 18, 2017

thelostone-mc commented Apr 18, 2017

thelostone-mc commented Apr 18, 2017

cjihrig commented Apr 18, 2017

thelostone-mc commented Apr 19, 2017

thefourtheye commented Apr 19, 2017

santigimeno commented Apr 19, 2017

thelostone-mc commented Apr 19, 2017 • edited Loading

santigimeno commented Apr 19, 2017

jasnell commented Apr 19, 2017

thelostone-mc commented Apr 19, 2017

santigimeno commented Apr 19, 2017

thelostone-mc commented Apr 20, 2017 • edited Loading

addaleax commented May 6, 2017

santigimeno commented May 6, 2017 • edited Loading

thelostone-mc commented May 6, 2017

thelostone-mc commented May 7, 2017 • edited Loading

gibfahn commented May 7, 2017

santigimeno commented May 7, 2017

thelostone-mc commented May 7, 2017

santigimeno commented May 7, 2017

thelostone-mc commented May 8, 2017

thefourtheye May 8, 2017

Choose a reason for hiding this comment

santigimeno May 8, 2017 • edited Loading

Choose a reason for hiding this comment

thefourtheye May 8, 2017

Choose a reason for hiding this comment

santigimeno May 8, 2017

Choose a reason for hiding this comment

thelostone-mc May 8, 2017

Choose a reason for hiding this comment

thefourtheye May 8, 2017

Choose a reason for hiding this comment

thelostone-mc commented May 8, 2017

thefourtheye commented May 9, 2017

jasnell commented May 9, 2017

Remove common.PORT usage from `parallel` tests #12473

Remove common.PORT usage from `parallel` tests #12473

thelostone-mc commented Apr 17, 2017 •

edited

Loading

thelostone-mc commented Apr 17, 2017 •

edited

Loading

thefourtheye commented Apr 18, 2017 •

edited

Loading

thelostone-mc commented Apr 19, 2017 •

edited

Loading

thelostone-mc commented Apr 20, 2017 •

edited

Loading

santigimeno commented May 6, 2017 •

edited

Loading

thelostone-mc commented May 7, 2017 •

edited

Loading

santigimeno May 8, 2017 •

edited

Loading